Skip to main content

Full text of "On the K- and L-theory of hyperbolic and virtually finitely generated abelian groups"

See other formats


(N 



in 






On the K- and L-theory of hyperbolic and 
virtually finitely generated abelian groups 



o 

(N 

^- Wolfgang Liick* 

^ '. http://www.him.uni-bonn.de/lueck 

Mathematisches Institut 

Universitat Bonn 

Endenicher Allee 60 

53115 Bonn 

f-H I Germany 

David Rosenthal^ 

Department of Mathematics and Computer Science 

St. John's University 

8000 Utopia Pkwy 

Jamaica, NY 11439 

<^: USA 

)Q. : May 2012 

OO 

o 
l> 

("^ ■ Abstract 

^-^ ■ We investigate the algebraic K- and L-theory of the group ring RG, 

where G is a hyperbolic or virtually finitely generated abelian group and 
R is an associative ring with unit. 

Sr^ ■ Key words: Algebraic K- and L-theory, hyperbolic groups, virtually finitely 

$H I generated abelian groups. 

•^ ! Mathematics Subject Classification 2000: f9D99, f9G24, f9A3f , f9B28. 

Introduction 

The goal of this paper is to compute the algebraic K- and L-groups of group 
rings RG, where G is a hyperbolic group or a virtually finitely generated abelian 
group and R is an associative ring with unit (and involution) . 



* email: wolfgang.lueck@him.uni-bonn.de 
^email: rosenthd@stjohns.edu 



The Farrell- Jones Conjecture for a group G and a ring R predicts that the 
assembly maps 

H^ {KG:. K,,) ^ H^ {G/G; K^) = K,, (RG) ; 

induced by the projection EG — > G/G, are bijective for every integer n. This 
conjecture, introduced by Farrell and Jones in their groundbreaking paper |21j . 
has many consequences. Knowing that it is true for a given group implies several 
other well-known conjectures for that group, such as the ones due to Bass, Borel, 
Kadison and Novikov. The Farrell- Jones Conjecture also helps to calculate the 
K- and L-theory of group rings, since homology groups are equipped with tools 
such as spectral sequences that can simplify computations. 

Recently it has been established that the Farrell- Jones Conjecture is true for 
word hyperbolic groups and virtually finitely generated abelian groups. Using 
this fact, we are able to compute the K- and L-thcory of their group rings 
in several cases by analyzing the left-hand side of the assembly map. The 
key ingredients used to compute these groups are the induction structure that 
equivariant homology theories possess [IHl Section 1] and the work of Liick- 
Weiermann J3T] , which investigates when a universal space for a given group G 
and a given family of subgroups J- can be constructed from a universal space 
for G and a smaller family T' C T. 

Even in basic situations determining the K- and L-groups is difficult. How- 
ever, we are able to handle hyperbolic groups. The favorite situation is when 
the hyperbolic group G is torsion- free and R — 'Z, in which case Kn (ZG) = for 
n < — 1, the reduced projective class group KqCEG) and the Whitehead group 
Wh(G) vanish, and K^i'^G) is computed by Hn{BG;K{Z)), i.e., the homol- 
ogy with coefficients in the X-theory spectrum K(Z) of the classifying space 
BG. Moreover, the L-theory L„ ' (ZG) is independent of the decoration (i) and 
is given by iJ„(i3G; L(Z)), where L(Z) is the (periodic) L-theory spectrum of 
L(Z). RecaU that 7r„(L(Z)) is Z if n ee mod 4, Z/2 if if n = 2 mod 4, and 
vanishes otherwise. We also give a complete answer for G = Z''. However, 
for a group G appearing in an exact sequence 1 ^ Z'^ ^- G ^- Q ^- 1 ior a 
finite group Q, our computations can only be carried out under the additional 
assumption that the conjugation action of Q on Z** is free away from or that 
Q is cyclic of prime order. 

In Section [T] precise statements of our results are given, as well as several 
examples. Section [2] contains the necessary background for the proofs, which 
are presented in Section [31 

The paper was supported by the Sonderforschungsbereich SFB 878 - Groups, 
Geometry and Actions -, the Leibniz-Preis of the first author, and the Fulbright 
Scholars award of the second author. The authors thank the referee for several 
useful comments. 



Contents 



Statement of Results 

1.1 Hyperbolic groups 

1.2 Torsion- free hyperbolic groups 

1.3 Finitely generated free abelian groups 

1.4 Virtually finitely generated abelian groups 

1.4.1 iT-theory in the case of a free conjugation action 

1.4.2 i-thcory in the case of a free conjugation action 

1.4.3 iiT-theory in the case Q — "L/p for a prime p and regular R 

1.4.4 i-theory in the case Q = Z/p for an odd prime p llO 



Background |ll 

2.1 Classifying Spaces for Families 11 

2.2 Review of the Farrell- Jones Conjecture 12 

2.3 Equivariant homology and relative assembly maps [14 



3 Proofs of Results 

3.1 Hyperbolic groups 

3.2 K- and L-theory of Z'' 

3.3 Virtually finitely generated abelian groups 

3.3.1 if -theory in the case of a free conjugation action 

3.3.2 L-theory in the case of a free conjugation action 

3.3.3 if-theory in the case Q — Z/p for a prime p and regular R 

3.3.4 i-theory in the case Q = Z/p for an odd prime p 

1 Statement of Results 

Let Kn{RG) denote the algebraic K -groups of the group ring RG in the sense of 
Quillen for n > and in the sense of Bass for n < —1, and let Ln {RG) be the 
ultimate lower quadratic L- groups oi RG (see Ranicki |37)). When considering 
L-theory, we will tacitly assume that i? is a ring with an involution. Let NKn{R) 
denote the Bass-Nil-groups of R, which arc defined as the cokernel of the map 
Kn{R) — > Kn{R[x]). Recall that the Bass-Heller-Swan decomposition says 

KniR'L) = K,,{R)(BK,,_iiR)(SNK„iR)®NK^iR). (1.1) 

If i? is a regular ring, then NKn{R) = for every n g Z (see for instance 
Rosenberg (351 Theorems 3.3.3 and 5.3.30]). 

The n-th Whitehead group Wh„(G';i?) is defined as 

WK{G;R) := H^ {EG ^ {•};Kn), 

where H^(^EG — > {•};Ki,;) is the relative term in the equivariant homology 
theory H^{~;K.ji] associated to the equivariant map EG -^ {•} (see 



Thus, Wh„(G'; R) fits into a long exact sequence 

> Hn{BG; K{R)) -^ Kn{RG) ^ Wli„(G; R) 

^ H„-i{BG;K{R)) ^ i^„-i(i?G) ^ • • • , 

where K(i?) is the non-connective iiT-theory spectrum associated to R and 
H^, (— ; K(i?)) is the associated homology theory. Notice that Whi(G'; Z) agrees 
with the classical Whitehead group Wh(G'). Whitehead groups arise natu- 
rally when studying /i-cobordisms, pseudoisotopy, and Waldhausen's A-theory. 
Their geometric significance can be reviewed, for example, in Dwyer- Weiss- 
Williams [20l Section 9] and Liick- Reich ^9] Section 1.4.1], where additional 
references can also be found. When G = Z, it follows from (|l.ll) and the fact 
that H^[EZ; Kr) = Kn{R) ® Kn~i{R) that there is an identification 

H^{EZ ^ {,};Kr) - NKn{R)®NKn{R). (1.2) 

Define the periodic n-th structure group with decoration (— oo) to be 

Sfr^-°^HG;R) ■■= i/„^(i?G^{.};4-°°>). 

These groups fit into the periodic version of the long exact surgery sequence 
with decoration (— cxd), 

> Hn{BG;L^-°°> (R)) -^Li-°^^{RG) -^ SP'''^^-°°>{G;R) 

^ ff„_i(i?G;L<-°°>(i?)) -> Li-T^(i?G) ^ • • • , 

where L^~°°' (R) is the spectrum whose homotopy groups are the ultimate lower 
quadratic L-groups. This periodic surgery sequence (with a different decoration) 
for i? = Z appears in the classification of ANR-homology manifolds in Bryant- 
Ferry- Mio- Weinberger [9J Main Theorem] . It is related to the algebraic surgery 
exact sequence and thus to the classical surgery sequence (see Ranicki [5S1 Sec- 
tion 18]). For n e Z define 

Vmii-^\D^;R) := H,f- (^i?oo ^ {•}; 47°°>). (1.3) 

These groups are related to Cappell's UNil-groups, as explained in Section [5] 
Now we are able to state our main results. 

1.1 Hyperbolic groups 

Theorem 1.4 (Hyperbolic groups). Let G be a hyperbolic group in the sense of 
Gromov '24 1, and let Ai be a complete system of representatives of the conjugacy 
classes of maximal infinite virtually cyclic subgroups ofG. 

(i) For each n Eli there is an isomorphism 

H^{EG;Kr) ® hX{EV ^ {•};Kr) ^ Kn{RG)- 
veM 



(ii) For each n G Z there is an isomorphism 

veM 

provided that there exists riQ < —2 such that Kn{RV) = holds for all 
n < hq and all virtually cyclic subgroups V C_ G. (The latter condition is 
satisfied if R — T, or if R is regular with Q C R.) 

A good model for EG is given by the Rips complex of G (see Meintrup- 
Schick [32^). Tools for computing H^ (^E_G;K.]ij are the equivariant version of 
the Atiyah-Hirzebruch spectral sequence (see Davis-Liick [161 Theorem 4.7]), the 
p-chain spectral sequence (see Davis-Liick J17') and equivariant Chern characters 
(see Liick [H]). More information about the groups H^ [EV — >■ {•};K/j) and 

HX{EV -^ {•}; L^"°°^) is given in SectionH 

1.2 Torsion- free hyperbolic groups 

The situation simplifies if G is assumed to be a torsion-free hyperbolic group. 

Theorem 1.5 (Torsion- free hyperbolic groups). Let G be a torsion-free hyper- 
bolic group, and let A4 be a complete system of representatives of the conjugacy 
classes of maximal infinite cyclic subgroups ofG. 

(i) For each n E Z there is an isomorphism 

H,,{BG;K{R)) (B ^ NK„{R) ® NK^{R) ^ K„{RG); 

VeM 

(ii) For each n G Z there is an isomorphism 

i/„(SG;L<-°°>(i?)) A lI-°"\rG). 

In particular, it follows that for a torsion-free hyperbolic group G and a 
regular ring R, 

KniRG) = {0} for n < 1 

and the obvious map 

Ko{R) ^ Ko{RG) 
is bijective. Moreover, Kq{ZG), Wh(G') and ii'„(ZG) for n < — 1 all vanish. 

Example 1.6 (Finitely generated free groups). Let Fr be the finitely generated 
free group *[^x^ ^f rank r. Since F,. acts freely on a tree it is hyperbolic. By 
Theorem [LSI 



Kn{RFr) = KniR) (B Kn^iiRY (B (NKniR) (B NK^R)) 

veM 

and 

Li-°°HRFr) = Li-^HR)®LtfiRy, 

where 7W is a complete system of representatives of the conjugacy classes of 
maximal infinite cyclic subgroups of Fr . 



1.3 Finitely generated free abelian groups 

Before tackling the virtually finitely generated abelian case, we first consider 
finitely generated free abelian groups. The iiT-theory case of Theorem 11.71 below 
is also proved in ^^. 

Theorem 1.7 (Z''). Let d > 1 be an integer. Let A4XCy be the set of maximal 
infinite cyclic subgroups of Z . Then there are isomorphisms 

Wh„(Z'*;i?) = ^{NKn-^{R) (B NK,,^,{R)y ^ > ; 
ceMicy 1=0 

K^iRiZ"]) = f0if„_,(i?)(')J ®Wh„(Z'^;i?); 

d 



Lt°^\R[Z'^]) = ^LtrHRp- 



i=0 
yd 



Example 1.8 {Z'^ x G). Let G be a group. By Theorem[L7] 

Kn{R[G xZ''])= KJRG[Z'']) 



d-l 



Kn-^{RGf^) © 0(Mf„_,(i?G) © NKn-^iRG)) 

1=0 ceMxcy(Z'^) i=o 



i"-') 



where A4ICy{Z'^) is the set of maximal infinite cyclic subgroups of Z"*. Since 
i/„ (B(G X Z-^); K(i?)) - iJ„ (BG; K(i?)) ^-^ , 

1=0 

this implies 

d 

Wh„(GxZ'^;i?) = 0Wh„_,(G;i?)(-) 



i=0 

d-1 

® 0(iVX„-^(i?G)®iVX„_,(i?G))' 

ceMicyiZ'^) i=o 



Example 1.9 (Surface groups). Let Tg be the fundamental group of the ori- 
entable closed surface of genus g, and let A1 be a complete system of represen- 
tatives of the conjugacy classes of maximal infinite cyclic subgroups of G. If 
g = 0, then Fg is trivial. If g = 1, then F^ is Z^ and Theorem 11.71 implies 



K„iRTi) = K„iR) © K^-iiRf © i^„-2(i?) © {NK,,{Rf © NK,,^i{Rf) 

veM 

and 

4— )(i?Fi) = Li— )(i?)©Li-lf (i?)2©4-_~>(i?). 



If g > 2, then Tg is hyperbolic and torsion-free, so by Theorem 1 1.5 1 

Kr^iRTg) ^ H„{Brg;KiR))S) NK^iRf, 

veM 

and 

Since Tg is stably a product of spheres, 

i/„(Sr<,;K(i?)) - X„(i?)©if„_i(i?fs©i^„_2(i?), 
and 

iJ„(i3r,;L<— >(i?)) - 4— )(i?)©4-_^>(i?)2^©Lt^>(i?). 

If it! = Z, then for every i e {1, 0, -1, . . .} II {-oo}, il'^(Z) is Z if n = 
mod 4, Z/2 if n = 2 mod 4, and is trivial otherwise. Therefore, 

'Z©Z/2 ifn = 0,2 mod 4; 
L<;> (ZFg) = <^ Z9 if n = 1 mod 4; 

(Z/2)9 ifriEES mod 4. 

More generally, cocompact planar groups (sometimes called cocompact non- 
Euclidean crystallographic groups) , e.g. , cocompact Fuchsian groups, are treated 
in Liick-Stamm [30j for R = Z,. These computations can be carried over to 
arbitrary R. 

1.4 Virtually finitely generated abelian groups 

Consider the group extension 

l^A^G^Q^l, (1.10) 

where A is isomorphic to Z"* for some d > and Q is a finite group. The 
conjugation action of G on the normal abelian subgroup A induces an action of 
Q on A via a group homomorphism which we denote by p: Q — > aut(A). 

Let M.ICy{A) be the set of maximal infinite cyclic subgroups of A. Since 
any automorphism of A sends a maximal infinite cyclic subgroup to a maximal 
infinite cyclic subgroup, p induces a Q-action on j\4ICy{A). Fix a subset 

/ C Micy{A) 

such that the intersection of every Q-orbit in AiICy{A) with / consists of 
precisely one element. 
For C e /, let 

Qc ^ Q 



be the isotropy group of C G AHCy{A) under the Q-action. Let 

h - {C E I \Qc = {!}}, 
h = {CeI\Qc = '1/2), 

and let J be a complete system of representatives of maximal non-trivial finite 
subgroups of G. 

1.4.1 iiT-theory in the case of a free conjugation action 

Theorem 1.11. Consider the group extension 1— s>y4— S'G— >Q— )-l, where A 
is isomorphic to Tfl for some d > and Q is a finite group. Suppose that the 
Q-action on A is free away from £ A. 

(i) For each n G Z, there is an isomorphism induced by the various inclusions 
( Wh„(^; i?) J ® (Z ®ZQ Wh„(A; R)) ^ Wh„(G; R). 

(ii) For each integer n, Z (XizQ Wh„(A; R) is isomorphic to 

Vce/i 1=0 J Xceh i=o J 

If Q = {1}, then Theorem [TTT] reduces to Theorem [TTl since J = 0, /2 = 
and /i is the set of maximal infinite cyclic subgroups of G = Z*^ . 

Theorem 1.12. Under the assumptions of Theorem ] 1 . 11[ 

(i) If R is regular, then for each n G Z 

0Wh„(^;i?)-Wh„(G;i?). 

In particular, 

^K„{RF) = K„ (RG) for n < -I 

FGJ 

and 

coker(i^o(i?) -^ Ko{RF)) = coker(ii'o(i?) ^ Ko{RG)). 
Fej 

(ii) If R is a Dedekind ring of characteristic zero, then 

Kn{RG)^{0} forn<~2. 



Applying Theorem 11.121 to the special case R='L,we recover the fact that 
if G satisfies the assumptions of Theorem 11.111 then: 



Wh(F) ^ Wh(G); 
FeJ 

0Xo(ZF) ^ Ko{1G)- 
FeJ 

0X_i(ZF) ^ X-i(ZG); 
Fe.J 

Kn{ZG) ^ {0} forn<-2. 

1.4.2 L-theory in the case of a free conjugation action 

Theorem 1.13. Consider the group extension l^y4— ^G— s^Q— >1, where 
A is isomorphic to ifi for some d > and Q is a finite group. Suppose that 
the Q-action on A is free away from G A. Assume that there exists ng < —2 
such that Kn{RV) — for all n < uq and all virtually cyclic subgroups V ^ G. 



(By Theorem \1.12\ (ii) , this condition is satisfied if R is a Dedekind ring of 



characteristic zero). 

Then there is an isomorphism 



(0Sr'<-°°>(F;i?))®(0 ljm\i-°°\D^;R)\ ^ 5r^<— >(G; i?), 
\FeJ ) XcehHeJc ) 

where Jc is a complete system of representatives of the conjugacy classes of 
maximal finite subgroups of WqC ~ NqC/G. 
If Q has odd order, then 

^r'^^"""^ (F; R) ^ 5^''^"°°^ (G; R). 
FeJ 

Dealing with groups extensions of the type described above when the con- 
jugation action of Q on A is not free is considerably harder than the free case. 
However, we are able to say something when Q is a cyclic group of prime order 
and R is regular. 

1.4.3 /-iT-theory in the case Q = Z/p for a prime p and regular R 

Theorem 1.14. Consider the group extension 1 ^> A — > G — > Z,/p — ;> 1, where 
A is isomorphic to TJ^ for some d > and p is a prime number. Let e be the 
natural number given by A^'P = 'If' , J be a complete system of representatives 
of the conjugacy classes of non-trivial finite subgroups of G, and A4ICy{A'^'P) 
be the set of maximal infinite cyclic subgroups of A^'P. 



// R is regular, then there is an isomorphism 
Wh„ (G; i?) - ( Wh„_, (ff ; i?) (°) ) 



\HeJ 4=0 

e-l 



0(M^„-»(i?[Z/p])©7Vif„_,(i?[Z/p])) 



(=7^) 



Remark 1.15 (Cardinality of J). Assume that the group G appearing in The- 
orem [TTTJ] is not torsion- free, or, equivalently, that J is non-empty. Consider A 
as a Z[Z/p]-niodule via the conjugation action p: "L/p — > aut(^). Then there is 
a bijection 

H^Z/p-A) A J, 

defined as follows. Fix an element f e G of order p. Every x £ iJ^(Z/p; A) is 
represented by an element x in the kernel of X]r=o P^ '■ ^ ~^ ^- Then xt has 
order p. Send x to the unique element of J that is conjugate to {xt) in G. 

If p is prime and R is regular, then Example 11.81 in the case G = Z/p is 
consistent with Theorem 11.141 and Remark 11.151 Namely, Z/p acts trivially on 
Z^, and so A = A^/P, J = {{0} x Z/p}, and d = e. 

1.4.4 L-theory in the case Q — Z/p for an odd prime p. 

Theorem 1.16. Consider the group extension 1 — s> A — ;■ G — )■ Z/p — s> 1, where 
A is isomorphic to Z'^ for some d > and p is an odd prime number. Let 
e be the natural number given by A '^ = Z'^ and J be a complete system of 
representatives of the conjugacy classes of non-trivial finite subgroups ofG. 
Then there is an isomorphism 

^^S^lt-'^H.Rp) ^Sr^^-^\G,R). 

HGJ i=0 

Two-dimensional crystallographic groups are treated in Pearson ^33] and 
Liick-Stamm [30) for R = Z. Some of the computations there can be carried over 
to arbitrary R. The Whitehead groups of three-dimensional crystallographic 
groups are computed in Alves-Ontaneda |ll. 

In the case R = Z, we get a computation for all decorations. 

Theorem 1.17. Consider the group extension 1 — s> A — ;■ G — )■ Z/p — s> 1, where 
A is isomorphic to Z'^ for some d > andp is an odd prime number. Let e be the 
natural number given by A^'P = Z'' , J be a complete system of representatives 
of the conjugacy classes of non-trivial finite subgroups ofG, and e be any of the 
decorations s, h, p or {{j) | j = 1, 0, —1, . . .} 11 {—00}. Define 

Sr^^G-Z) := H^^{EG^{.}-1.1). 
10 



Then there is an isomorphism 

Remark 1.18. In general, the structure sets Sll'^f{H;7j) depend on the deco- 
ration e. We mention without proof that for an odd prime p 

NO} n odd. 

Remark 1.19. The computation of the structure set S^"'^{G;Z) when the 
conjugation action of Z/p on Z'^ is free plays a role in a forthcoming paper by 
Davis and Liick, in which this case is further analyzed to compute the geometric 
structure sets of certain manifolds that occur as total spaces of a bundle over 
lens spaces with d-dimensional tori as fibers. 

Remark 1.20 (Topological iiT-theory of reduced group C*-algebras). All of the 
above computations also apply to the topological iT-theory of the reduced group 
C*-algebra, since the Baum-Connes Conjecture is true for these groups. In the 
Baum-Connes setting the situation simplifies considerably because one works 
with EG instead of EG and hence there are no Nil-phenomena. For instance, 
if G is a hyperbolic group, then there is an isomorphism 

K^{EG)^KUC:.{G)) 

from the equivariant topological i^-theory of EG to the topological iiT-theory 
of the reduced group C*-algebra G*{G). In the case of a torsion- free hyperbolic 
group, this reduces to an isomorphism 

X„(BG)^if„(C;(G)). 

The case G = Z'^ Xp Z/p for a free conjugation action has been carried out for 
both complex and real topological X-theory in detail in [l^ Theorem 0.3 and 
Theorem 0.6]. 

2 Background 

In this section we give some background about the Farrell- Jones Conjecture and 
related topics. 

2.1 Classifying Spaces for Families 

Let G be a group. A family of subgroups of G is a collection of subgroups that 
is closed under conjugation and taking subgroups. Examples of such families 



11 



are 

{1} — {trivial subgroup}; 
TXM = {finite subgroups}; 
VCy = {virtually cyclic subgroups}; 
J^CL = {all subgroups}. 

Let J^ be a family of subgroups of G. A model for the universal space Ej:{G) 
for -F is a G-Ciy-complex X with isotropy groups in F such that for any G- 
CM^-complex Y with isotropy groups in F there exists a G-map Y ^ X that 
is unique up to G-homotopy. In other words, X is a terminal object in the 
G-homotopy category of G-GW^-complexes whose isotropy groups belong to T. 
In particular, any two models for Ejr[G) are G-homotopy equivalent, and for 
two families J-q C J^i, there is precisely one G-map Ejrg{G) — !> Ejr^{G) up to 
G-homotopy. 

For every group G and every family of subgroups J- there exists a model 
for Ejr{G). A G-GW^-complex X is a model for Ejr{G) if and only if the in- 
fixed point set X^ is contractible for every H va T and is empty otherwise. 
For example, a model for Eji£c{G) is G/G — {•}, and a model for £'|i].(G) is 
the same as a model for £'G, the total space of the universal G-principal bundle 
EG — > BG. The universal G-CW -complex for proper G-actions, Ejrxj\f{G), will 
be denoted by EG, and the universal space Evcy{G) for VCy will be denoted 
by EG. For more information on classifying spaces the reader is referred to the 
survey article by Liick [28] . 

2.2 Review of the Farrell-Jones Conjecture 

Let Hi be an equivariant homology theory in the sense of Liick |26[ Section 1] . 
Then, for every group G and every G-GM^-pair (AT, A) there is a Z-graded 
abelian group T-L'^{X, A), and subsequently a G- homology theory T-L'^ . For every 
group homomorphism a: H ^^ G, every iJ-GM^-pair (AT, A) and every n G Z, 
there is a natural homomorphism inda : H^f {X, A) — > H^(G Xq (A", A)), known 
as the induction homomorphism. If the kernel of a operates relative freely on 
(X, A) , then inda is an isomorphism. 

Our main examples are the equivariant homology theories iJ^(— ;Kij) and 
i?J(— ;L}j °° ) appearing in the AT-theoretic and L-theoretic Farrell-Jones Con- 
jectures, where R is an associative ring with unit (and involution) (see Liick- 
Reich [29l Section 6]). The basic property of these two equivariant homology 
theories is that 

HS{G/H;Ki,) = H,^{{.};Ki,) = K^{RH); 

HO(G/H;L^^->) = i/f({.};L<,— >) = Li—\RH), 

for every subgroup H <Z G. 



12 



The Farrell-Jones Conjecture for a group G and a ring R states that the 
assembly maps 

H^{KG;L^--'^) ^ H^{G/G;L^^-^)^Li—\RG\ 

induced by the projection EG — ?> G/G, are bijective for every n 6 Z [21]. The 
Farrell-Jones Conjecture has been studied extensively because of its geometric 
significance. It implies, in dimensions > 5, both the Novikov Conjecture about 
the homotopy invariance of higher signatures and the Borel Conjecture about 
the rigidity of manifolds with fundamental group G. It also implies other well- 
known conjectures, such as the ones due to Bass and Kadison. For a survey and 
applications of the Farrell-Jones Conjecture, see, for example, Liick-Reich [551 
Section 6] and Bartels-Liick- Reich [7]. 

Theorem 2.1 (Farrell-Jones Conjecture for hyperbolic groups and virtually 
Z'^-groups). The Farrell-Jones Conjecture is true if G is a hyperbolic group or 
a virtually finitely generated abelian group, and R is any ring. 

The i^-theoretic Farrell-Jones Conjecture with coefficients in any additive 
G-category for a hyperbolic group G was proved by Bartels-Reich-Liick [6]. 
The version of the Farrell-Jones Conjecture with coefficients in an additive 
category encompasses the version with rings as coefficients. The L-theoretic 
Farrell-Jones Conjecture with coefficients in any additive G-category for hyper- 
bolic groups and CAT(0)-groups was established by Bartels and Liick [5]. In 
that paper it is also shown that the X-theoretic assembly map is 1-connected 
for CAT(0)-groups. Note that a virtually finitely generated abelian group is 
CAT(O). Quinn [MJ Theorem 1.2.2]) proved that the iiT-theoretic assembly map 
for virtually finitely generated abelian groups is bijective for every integer n if i? 
is a commutative ring. However, the proof carries over to the non-commutative 
setting. 

Remark 2.2 (The Interplay of K- and L-Theory). L-theory L^^\rG) can 
have various decorations for i e {2, 1, 0, —1, —2, . . .} 11 {— cx)}. One also finds 
L%[RG) for e = p,h,s in the literature. The decoration (1) coincides with the 
decoration h, (0) with the decoration p, and (2) is related to the decoration 
s. For j < 1 there are forgetful maps Li^~^^' {R) — )■ Li^' (R) that fit into the 
so-called Rothenberg sequence (see Ranicki [35[ Proposition 1.10.1 on page 104], 
[371 17.2]) 

■ . • ^ L\t'\R) ^ LipiR) ^ H^\Z/2:K,iR)) 

->Ll\^>(i?)^L<£i(i?)^.... (2.3) 

i?"(Z/2; Kj{R)) denotes the Tate-cohomology of the group Z/2 with coefficients 
in the Z[Z/2]-module Kj{R). The involution on Kj{R) comes from the invo- 
lution on R. There is a similar sequence relating L^{RG) and L^-^{RG), where 
the third term is the Z/2- Tate-cohomology of Whf (G). 



13 



For geometric applications the most important case is i? = Z with the dec- 
oration s. If KqCZG), Wh(G) and KnCZG) for n < — 1 all vanish, then the 
Rothenberg sequence for n e Z implies the bijectivity of the natural maps 

L^(ZG) ^ L':,{ZG) ^ LliZG) ^ L^'^^m). 

In the formulation of the Farrell- Jones Conjecture (see Section [2]), one must 
use the decoration (—00) since the conjecture is false otherwise (see Farrell- 
Jones-Liick [25]V 

2.3 Equivariant homology and relative assembly maps 

Consider an equivariant homology theory H^ in the sense of |26[ Section 1]. 
Given a G-map / : X — > F of G-GW^-complexes and n G Z, define 

n^U) ■■= H^(cyl(/o),X) (2.4) 

for any cellular G-map Jo'- X -^ Y that is G-homotopic to /. Here, cyl(/o) is 
the G-GVF-complex given by the mapping cylinder of /. It contains X as a G- 
GW-subcomplex. Such an /o exists by the Equivariant Cellular Approximation 
Theorem (see [39i Theorem II. 2.1 on page 104]). The definition is independent 
of the choice of /o, since two cellular G-homotopic G-maps Jq, fi: X ^^ Y are 
cellularly G-homotopic by the Equivariant Cellular Approximation Theorem 
(see [Ml Theorem II. 2.1 on page 104]) and a cellular G-homotopy between /o 
and /i yields a G-homotopy equivalence u : cyl(/o) — ?> cyl(/i) that is the identity 
on X. This implies that H^^f) depends only on the G-homotopy class of /. 
From the axioms of an equivariant homology theory, H^(/) fits into a long exact 
sequence 

• • ■ ^ n^+iU) ^ nfAx) ^ HfAY) ^ n^if) ^ n^-,{x) -^ • • • . (2.5) 

The following fact is proved in B artels [3]. 
Lemma 2.6. For every group G, every ring R, and every n e Z.' 
(i) the relative assembly map 

H^{EG-Ki,) ^ H^{EG;Kr) 
is split-injective; 
(ii) the relative assembly map 

/f„^(^G;4-^>) ^F„«(^G;4-°°>) 

is split-injective, provided there is an uq < —2 such that Kn{RV) — for 
every n < uq and every virtually cyclic subgroup V Q G. 



Remark 2.7. Notice that the condition appearing in assertion (ii) of the above 



lemma is automatically satisfied if one of the following stronger conditions holds: 

14 



(i) i? = Z; 

(ii) R is regular with Q C R. 

li R = Z, this fohows from [H Theorem 2.1(a)]. If R is regular and Q C i?, 
then for every virtually cyclic subgroup V <^ G the ring RV is regular and hence 
K,,{RV) = for aU n < -1. 



Lemma 12.61 tells us that the source of the assembly map appearing in the 
Farrell- Jones Conjecture can be computed in two steps, the computation of 
Hn {EG; Kfl) and the computation of the remaining term H^ {EG -^ EG; Kj^) 
defined in (12. 4p . Furthermore, Lemma 12.61 implies 

H^ {EG ^ EP\ ^r) = H^ {EG -^ EG; Kr) ® H^ {EG -^ EG; Kr) ; (2.8) 

and 



H^{EG^KG;L^-°°^)^ 



H^{EG^EG;L^^:^^] 



Hl^{EG^EG;i^R 



(-oo>^ 



(2.9) 



provided that there exists an ng < —2 such that Kn{RV) = for every n < uq 
and every virtually cyclic subgroups y C G. This is useful for calculating the 
Whitehead groups for a given group G and ring R which satisfy the Farrell- Jones 
Conjecture. 

Remark 2.10. The induction structure of H^{—;K.ji) can be used to define a 
Z/2-action on H^{EZ -> {•};Kfl) that is compatible with the Z/2-action on 
NKn{R) ® NKn{R) given by flipping the two factors. The action is defined by 
the composition of isomorphisms 



T : Hl{EZ ^ {•};-Ki 



^iI^(ind_id,ra^{.};K;^) 

^ Hl{EZ ^ {•};-Kr) 



where the second map is induced by the unique (up to equivariant homotopy) 
equivariant map I : ind_ ids EZ — >■ EZ; I is an equivariant homotopy equivalence 
since ind_ id^ EZ is a model for EZ. 

To see that this corresponds to the flip action on NKn{R)®NKn{R), consider 
the following diagram coming from the long exact sequence (|2.5p . 



H^{EZ;Kr) 



-ff5(W;Kfl) 



Hf,{EZ ^ {.};Kr) 



ind- 



H,^(ind„id,ra;Kfl) 



H^{EZ;Ki 



ind_ id- 

^H^{{.};Kr) 
^Hf,{{,};KR) 



ind_ 



-4i/^(ind_id,ra^{.};Kji) 



-^ H^{EZ ^ {.};Kr) 



15 



Recall that H^{EZ;Kb) -^ H^{{»};Kr) ^ Kn{R'Z) = Kn{R[t,t-^]) is a 
split injection, and so the Bass-Heller-Swan decomposition (|l.ip establishes the 
identification H^{EZ -> {•}; Kr) ^ NK„iR) © NK„{R) dO]). Also recall that 
the two copies of NKn{R) appearing in the decomposition of Kn{R[t, t~^]) come 
from the embeddings R[t] ^^ R[t,t~^] and R[t~'^] ^-> R[t,t~^]. By the definition 
of the induction structure for Hl{—; Kr), 

ind_id, : H^{{>};Kr) ^ i/,f({.};Kfl) 

corresponds to the homomorphism Kn{R[t,t~^]) — J> Kn{R[t,t~^]) induced by 
interchanging t and t~^ (see, for example, [5S1 Section 6]), which swaps the two 
copies of NKn{R) in the decomposition of Kn{R[t,t~^]). Therefore the above 
diagram implies that r coincides with the flip action on NKn{R) ® NKn{R). A 
proof of this fact can also be found in [191 Lemma 3.22]. 

Remark 2.11 (Relative assembly and Nil-terms). Let V be an infinite virtually 
cyclic group. If V is of type I, then V can be written as a semi-direct product 
F X Z, and HY (^EV — > {•};K.r) can be identified with the non-connective 
version of Waldhausen's Nil-term associated to this semi-direct product (see [H 
Sections 9 and 10]). If V is of type II, then it can be written as an amalgamated 
product Vi *vb V2 of finite groups, where Vq has index two in both Vi and ¥2- 
In this case, HY (^EV — > {•};Ki(;) can be identified with the non-connective 
version of Waldhausen's Nil-term associated to this amalgamated product [4] . 
The identifications come from the Five-Lemma and the fact that both groups 
fit into the same long exact sequence associated to the semi-direct product, 
or amalgamated product, respectively. This is analogous to the i-theory case 
which is explained below. If R is regular and Q Q R, e.g., 7? is a field of 
characteristic zero, then H^ {EV — >■ {•};K/j) = for every virtually cyclic 
group V, and hence, for any group G the map 

H^{EG;Kr) ^ H^{EP;Kr) 

is bijective (see [29j Proposition 2.6]). 

Let UNil„(i?; i?, i?) denote the Cappell UNil-groups fTO] associated to the 
amalgamated product Z?oo — Z/2 * Z/2. It is a direct summand in L'^{R[Doo]) 
and there is a Mayer- Vietoris sequence 

> Ll{R) ^ L';^{R[Z/2]) © Lf,(ii'[Z/2]) ^ L';^{R[Do^])/ UNil„(i?; R, R) 

^ L:i(i?) ^ L':,iR[Z/2]) ® 4— >(i?[Z/2]) ^ . • • . 

There is also a Mayer- Vietoris sequence that maps to the one above which comes 
from the model for EJDoo given by the obvious _Doo-action on M: 

• • • ^ L':^ ^ Lt{R[Z/2]) © Lt{R[Z/2]) ^ H^- {ED^;L%) 

-> Li\{R) ^ L'^{R[Z/2]) © L'^mm]) ^ • ■ • ■ 



16 



This implies 

Lemma 2.12. Assume that Ki{R) ^ Ki{R[I./2]) = K,,{R[Doo]) = for all 
i <0. (This condition is satisfied, for example, if R. = Z.) Then: 

UNil,'j(i?; R, R) ^ UNil<-°°> {Doo] R)- 
Proof. Using the Rothenberg sequences (12.31) , one obtains natural isomorphisms 

4— >(i?) - Lf,(i?); 

Thus, a comparison argument involving the Atiyah-Hirzebruch spectral sequences 
shows that the obvious map 

is bijective for all n G Z. D 

Cappell's UNil-terms [10] have been further investigated in [2],[12] and [13]. 
The Waldhausen Nil-terms have been analyzed in [15] and [25]. If 2 is inverted, 
the situation in i-theory simplifies. Namely, for every n 6 Z and every virtually 
cyclic group V, H^ {EV -^ {•}; LJj°°^)[l/2] = 0. Therefore the map 

i/„«(^G;4-°°>)[l/2] A i?,<r(^G;4-°°>)[l/2] 

is an isomorphism for any group G, and the decorations do not play a role 
(see [m Proposition 2.10]). 

Remark 2.13 (Role of type I and II). Let VCyi be the family of subgroups 
that are either finite or infinite virtually cyclic of type I, i.e., groups admitting an 
epimorphism onto Z with finite kernel. Then the following maps are bijections 

Hn{Evcyi{G);'KB) ^ if„(^G;Kij); 

Hn (KG'i'L'fi °° ) ^ i?„ (EvcyiiG); L^j °° )■ 

For the iiT-theory case, see, for instance, [El [19]. The L-theory case is proven 
in Liick [27| Lemma 4.2]). In particular, for a torsion-free group G, the map 

i7„«(i?G;L<,— >) ^ i/„«(^G;L<,— >) 
is a bijection. 



17 



3 Proofs of Results 

We now prove the results stated in Section [TJ 

3.1 Hyperbolic groups 

Proof of Theorem \1.4\ By [31, Corollary 2.11, Theorem 3.1 and Example 3.5] 
there is a G-pushout 

UveMG>^vEV^-^EG 

UveMP 

UveMG/V >MG 

where i is an inclusion of G-CW^-complexes, p is the obvious projection and A4 is 
a complete system of representatives of the conjugacy classes of maximal infinite 
virtually cyclic subgroups of G. Now Theorem 11.41 follows from Theorem 12.11 
and Lemma [2.61 D 



Proof of Theorem \1.5[ (i) This follows from (|1.2I) and Theorem 11.41 (i) 



Since any virtually cyclic subgroup of G is trivial or infinite cyclic, the claim 
follows from Theorem 12.11 and Remark 12.131 D 

3.2 K- and L-theory of Z'^ 

As a warm-up for virtually free abelian groups, we compute the K and L-theory 

of i?[Z'']. 

Proof of Theorem \1.7\ Using the induction structure of the equivariant homol- 
ogy theory _ffj(— ;K/;) and the fact that B'Z'^ is the d-dimensional torus, it 
follows that 



d 

0K„_,(i?)('). (3.1) 



4=0 

Similarly one shows that 



Hf{EZ'^;L^--^) = 04-f(i?)(^). (3.2) 

i=0 

Theorem 12. II implies that 

WKiZ'^;R) := Hi' {EZ'^ -^ {•};Kr) = H^^EZ'^ ^ £Z'^,Kr). 



18 



From [211 Corollary 2.10] it follows that 

Hf {EZ'^ -^ EZ'^/C- Kr) = H^' {EZ'^ -^ ^Z'^; Kr). 



ceMicy 



Since C CZ'^ is maximal infinite cyclic, Z'* = C®Z'^ ^. Therefore the induction 
structure and (|1.2p imply 



Hi {EZ'' -^ E{Z'^/C)-Kr) ^ H^®^ {EC x EZ''-^ -^ EZ'^-^-^Kr) 

^ H^{EC xBZ"^-^ ^ BZ^'-^-Kr) 

^ H^{iEC^{»})xBZ'^-^-KR) 

- '^HS_,{EC^{.}:KRf''^ 



i=0 
d-1 



^ 0(Mf„_,(i?)©Mr„_,(i?))^ ' ^ (3.3) 



Hence, 



d-l 



Wh„(Z'^;i?) - 0(iVK„_,(i?)©M^„_,(i?)) ('-'). 



CGA4ICJ' 1=0 



From Theorem 12.11 and Lemma 12.61 (i) 

KniR[Z'']) = i/f (ra'*;K^)©Wh„(Z'^;i?), 
which, by p.ip . is isomorphic to 

d 



0i^„_,(7?)(') ©Wh„(Z'';i?). 



\i=0 

Finally, 



Li'^HRi^']) ^ i?r(ra'^;4— >) - 0i<-_f (i?)(?) 



i=0 



by Theorem O Remark [^1^ and ([22]) . D 

3.3 Virtually finitely generated abelian groups 

For the remainder of the paper we will use the notation introduced in Subsec- 
tion 11.41 and the following notation. For C E I there is an obvious extension 

1 ^ C ^ NgC ^ WgC -^ 1, 



19 



where pc is the canonical projection. We also have the extension 

l^A/C^ WgC ^Qc^l, (3.4) 

which is induced by the given extension 1— >j4— >G— >Q^1. Since C C A is 
a maximal infinite cyclic subgroup, A/C = TU^^^ . 

Notice that any infinite cyclic subgroup C of A is contained in a unique 
maximal infinite cyclic subgroup Cmax of A. In particular for two maximal 
infinite cyclic subgroups C^D C A, either Cnl? = {0} or C = I?. Let 

iVG[C] := {.g e G I IgCg-^nCI-oo}. 

For every C & I 

NG[C]^NGC^q-\Qc). 

Consider the following equivalence relation on the set of infinite virtually cyclic 
subgroups of G. We call Vi and V2 equivalent if and only if {A n Vi)i„ax = 
{A n V2)max- Then for every infinite virtually cyclic subgroup ^ of G there is 
precisely one C € I such that V is equivalent to gCg^^, for some g G G. We 
obtain from [ST] Theorem 2.3] isomorphisms 

^H^^''{ENgC-^P*cEWgG;Kr) - H^{EG ^ EP;Kr); (3.5) 
cei 

^H^^''{ENgC^P*gEWgG;L^^'^^) = H^ {EG ^ RG^L^^"^^), (3.6) 
cei 

where pc ■ NgC = q^^{Qc) -^ WgC ~ q^^{Qc)/G is the canonical projection. 

Lemma 3.7. Let /: Gi — > G2 be a surjective group homomorphism. Consider 
a subgroup _ff C G2. Let Y be a Gi-space and Z be an LI -space. Denote by 
fn- f~^{LL) — >■ LI the map induced by f. 

Then there is a natural Gi -homeomorphism 

Gi ^f-^iH) (res^r*""^ ^ ^ fnZ) ^ ^ ^ f*^^^ ^« ^)' 

where f^, f* and reSg denote restriction and the actions on products are 

the diagonal actions. 

Proof. The map sends {g,{y,z)^ to {gy , {f {g) , z)) . Its inverse sends (y, (fc, z)) 
to (/i, {h^^y, z)) for any /i G Gi with f{h) = k. D 

3.3.1 ii'-theory in the case of a free conjugation action 

Proof of Theorem \L11[ We prove assertions (i) and (ii) simultaneously by a 
direct computation. 

By Theorem [2l]i^^(lG;Kfl) = H^{{m};KR). Thus, 

Wh„(G;i?) := H^ {EG ^ {my^KB) 
^ H^ {EG ^ EP;Kh) 
= H^ {EG -^ EG; Kr) ® H^ {EG -^ EG; Kr) , 

20 



using (|2.8p . From f33 Lemma 6.3] and [3T1 Corollary 2.11], there is a G-pushout 

Ui.ejGxi.£;i^^^£;G (3.8) 

U^^^G/F >^G 

where J is a complete system of representatives of maximal finite subgroups of 
G. This produces an isomorphism 

WK{F;R) := <(i?F ^ {•};Kn) ^ H^ {EG ^ EG:Kj,) . 
FeJ FGJ 

BydSSI, 

H^''''{ENgC ^ p*cEWgC; K^) - i7,? {EG -> ^G; K^) . 

Fix G in /. Since the conjugation action of Q on A is free away from 0, 
it induces an embedding of Qc into aut(G). Hence Qc is either trivial or 
isomorphic to Z/2. Therefore, I — Ii 11/2, where /i = {G G / | Qc = {!}} a-nd 
/2 = {Gg/|Qc = Z/2}. 

From [31, Corollary 2.10] there is an isomorphism 

H^^^{ENaC ^ p*cEWaC; Kr) 

ceMicy(A) 

^ H^ {EA -^ EA; Kfl) = Wh„(A; R), 

where pc ■ NaC = A — > WaC = A/C denotes the projection. The conjugation 
action of Q on A induces an action of Q on H^{EA -^ EA-jK-j^j. By the 
definition of the index set / and the subgroup Qc C Q, we obtain a ZQ- 
isomorphism 

ZQ (^^Q^] H^{EA ^ p^jEiA/GY-Kn) - H^{EA ^ RA; K«), 

and hence an isomorphism 

Z ®z[Q^] i/^(i?A ^ p*cE{A/G)- Kr) - Z ®ZQ H^{EA ^ ^A; Kr). 
cei 

If Qc = {!}, then NcG = A = GeZ'^-i and WgG = Z'^-\ Thus, by (1531) . 

H^^^{ENgG-^P*cEWgG;Kr) = H^{EA ^ p*cE{A/C);Kr) 

^ H^®^"" {EC X EZ"-' ^ ra"*-!; Kfl) 

= ^{NKn-^{R) (B NK^^,{R)y ^ ' . 

i=0 



21 



Therefore, the proof of the theorem will be finished once it is established that 
for C e h, 

H^"^ {ENgC ^ p*cEWgC- Kr) 

d-l 

= Z ®j,^Q^^ H^{EA ^ p*cE{A/C); Kr) - 7Vi^„_,(i?)('"). 

Assume that Qc = Z/2. Since the projection ENqC x p*(jEWgC -^ ENqC 
is an A^cC-homotopy equivalence, H^'^^ (ENgC -^ p'^E_WgC;Kii) is isomor- 
phic to H^^^{ENgC X p*gEWgC -^ p*cEWgC; Kr) , which, by LemmaEHi) 
below, is isomorphic to 

H^^^{ENgC X pUEWgC X qU*EQc)-^phiEWGC x q^*EQc);KR). 

Sending a Qc-Ciy-complex Y to the Z-graded abelian group 

H^'^^{ENgC X p*ciEWGC X q^*Y) ^ p^EWgC x q^*Y);KR) 

yields a Qc-homology theory. Since EQg is a free Qc-C'W^-complex, there is 
an equivariant Atiyah-Hirzebruch spectral sequence converging to 

H^+f{ENGC X p*c{EWgC x qd*EQc) ^ p*c{EWgC x qd*EQc); Kr) 

whose _E^-term is given by 



Kj = ^i [EQc; Hf^^{ENGC X p*c{EWgC x q^*Qc) 

-^ pUEWgC xq^*Qc)]KR] 
Lemma I5^| (ii) | implies that: 



El^ ^ H^^{EQc;Hf{EA^p*GE{A/C);KR)) 

- iJ^- {EQc;nQc] ®z (Z®z[Qc] Hf{EA^p*GE{AIC)-KR))) 

= hP (res J] EQc; Z ^^[q^] Hf{EA ^ p*cE{AIC)- Kr)) 

^ {■L®^^Q^^Hf{EA^p*cE{A/C)-KR) i^Q 
\{0} ^^0. 

\{0} *^0. 

Hence, the Atiyah-Hirzebruch spectral sequence collapses at E"^ . Therefore, 
H^<^^{ENgC X p*c{EWgC X qd*EQc) ^ Pc(^P^gC x qd*EQc); Kk) 

d-l 

- Z ®^[Q^] i7;f (i?A ^ pt£;(A/C); Kr) - A^i^„_Ki?)('"). 

i=0 



This completes the proof of Theorem 11.111 once we have proved Lemma 1331 D 

22 



Lemma 3.9. Let G be a group satisfying the assumptions of Theorem ] 1 . 1 11 and 
let C be a maximal cyclic subgroup of G such that Qc = Z/2. Then: 

(i) For every n € Z, the projection EWqC x qc* EQc -^ EWqC induces a 
bijection, 

H^^''{ENGCxp*ciEWGCxq^*EQc)^P*c{EWGCxq^*EQc);KR) 
^ H^^'^iKNaC X p*cEWgC ^ p*gEWgC; Kr). 

(a) There are isomorphisms of 1i[QG]-m,odules 

Hf°^{ENGC X p*c[EWgC x qd*Qc) ^ pUEWgC x qd*Qc); K^) 

d-l 

- Hf{EA ^ p*GE{A/C)-Kn) - 0(Z[Qc] ®z ^i^j-K^)) 



z=o 



Proof, (i) Note that the conjugation action of Qc = Z/2 on A/C is free away 
from € ^/C. To see this, let g be a nontrivial element in Q, and let aC € A/C 
such that aC is fixed under the conjugation action with q. Then there is a c £ C 
such that p[q)[a) = a + c, where p: Q ^ aut(A) is given by the conjugation 
action of Q in A. This implies p{q){2a) = 2a + 2c. Thus, 

p{q){2a + c) = p(g)(2a) + p{q){c) = p{q){2a) - c = 2a + 2c - c ^ 2a + c. 

By assumption, the Q-action on A is free away from G A, so 2a + c = and 
thus 2a e C. Since C is maximal infinite cyclic, aC = in A/C. Hence, the 
conjugation action of Qc on A/G is free away from G A/G. 

Since (A/C)'^"^ = {0}, the extension p.4p has a section, and WgC* is iso- 
morphic to the semidirect product Z"^"^ x Z/2 with respect to the involution 
— id: Z''^^ — i> Z''"^. Let Jc be a complete system of representatives of the 
conjugacy classes of maximal finite subgroups of WgG. Then every element 
in Jc is isomorphic to Z/2 and Jc contains 2''"^ elements (see Remark II .ISp . 
From [Sni Lemma 6.3] and i31j Corollary 2.11], there is a VFcC'-pushout 

UHeJc^G'^''HEH^-^EWGG (3.10) 

UneJcP 

UneJc WgG/H > EWgG 

Let H & Jc. Given any iVcC-space Y and any i?-space Z, Lemma [3.71 
produces an A^gC-homeomorphism 

NgG Xp-.(^) (resg^'^^y) x p^Z) ^Y x p^WcG xh Z), (3.11) 



23 



where p}j denotes restriction for both pc'- NqC — ?> WqC and the homomor- 

phism Pc\H) -^ H induced by pc- Since res^^^J^lf ^^iVcC) is a p^\H)-CW- 
model for Ej)'^^{H), we obtain identifications of iVgC-spaces 

ENgC X pUWgC xhEH) ^ NgC Xp-i(^) {EPc^H) x p*cEH); 

ENgCxp*c{WgC/H) = NgCx^-i^^^{Ep-\H))- 

P*c{WgCxhEH) = NgCx^-h^^p*cEH; 

pUWgC/H) - A^gC x^-.(^) {.}, 



by substituting Y = ENgC or Y = {•}, and Z ^ EH or Z ^ {•} into (I5TT|) . 
Using the induction structure of the equivariant homology theory i?^(--;Kfl) 
and the Five-Lemma, these identifications imply 



H^''^ {ENgC X p^WgC xh EH) ^ p*c{WgC xh EH);Ki,) 



Hl^ ^"> {Ep-}{H) X p*cEH ^ p*cEH; Kr) (3.12) 



and 



-,\H) 



^ H^^^{ENgC X p*c{WgCIH)^p*c{WgC/H)-Kr) (3.13) 

are bijective for every n & "L. 

Consider the following p^"'^(iJ)-pushout, where the left vertical arrow is an 
inclusion of p^ (iJ)-CVF-complexes and the upper horizontal arrow is cellular. 

Epc\H) >p*cEH 



-^X 



El>c\H) 



Then X is a pp'^(i?)-CW^-complex, and for every subgroup K C Pf-,^{H), the 
above p^^(iJ)-pushout induces a pushout of CW^-complexes: 



Epc\H) 



K 



Epc\H) 



K 



-, {p},EH) 



K 



^X 



K 



If K = {1}, then the spaces Ep-\H)'^ , {p}.EH)^ and Ep-^{H)^ are con- 
tractible, and thus X^ is contractiblc. If iiT is a non-trivial finite subgroup, 
then the spaces Ep^^{H)^ and (p'^EH)^ are empty, and the space Ey'^^jH)^ 
is contractiblc. Hence X^ is contractiblc. If K is an infinite subgroup of C, 



24 



then the spaces Ep'^^{H)^ and EpQ^{H)^ are empty, and the space (p'^EH)^ 
is contractible since pc{K) = {!}• Therefore X^ is contractible. If K is 
infinite and not contained in C, then the spaces Ep'^^{H)^, {p'^EH)^ and 
Ep'^^{H)^ are all empty, and so X^ is also empty. Since p^^{H) = D^o, an 
infinite virtually cyclic subgroup K of p'^ (H) is of type I if and only if it is 
an infinite subgroup of C. Hence, X is a model for Eycyi{Pc^{H))- Notice 
that Ep^^{H) X PqEH is a model for Ep^^{H). Therefore, we have the fol- 
lowing cellular p^ (i?)-pushout, where the left vertical arrow is an inclusion of 
p^^(7f)-CW^-complexes and the upper horizontal arrow is cellular. 

Ejpc\H)xp*cEH >P*cEH 



Ep^\H) >EvcyAPdHH)) 

This induces an isomorphism 

if^''""^ {Epc\H) X p*cEH ^ p*cEH; Kj,) 

^ H^^'^"\Epc'{H) ^ EvcyAPd\H));Kn), (3.14) 

for every n G Z. Since p^ (H) = Doo is virtually cyclic. Remark 12.131 implies 
that the map 

H^„^'^"\EycyAPc'iH));^R) ^ Hr/'^"\{.};Kj,) 
is bijective for every n e Z, and so by p.l4p . 

H^^'^"^ {Ej>c\H) X p*cEH ^ p*cEH- ^r) 

^ H^y^"\Ep~\H) ^ {.};K«) (3.15) 
is bijective for every n e Z. Therefore, p.l2p . p.l3p and p.lSp imply that 

H^'^^{ENgC X p*c{WgC Xh EH) ^ p^WgC xh EH);Kr) 

^ H^^'^iENaC X p*c{WgC/H) ^ p*c{WgC/H);Kr) (3.16) 

is an isomorphism for every n € Z. 

We obtain a WgC'-homology theory by assigning to a W^gC'-CM^-complex 
Z the Z-graded abehan group H^^^ {ENgC x p^Z -^ pJ,Z;Kfl). From the 
Mayer- Vietoris sequence associated to the VFcC-pushovit (13.101) and the bijec- 
tivity of the map p.l6p , there is an isomorphism 

H^""" {ENgC X p*cEWgC ^ p*cEWgC; Kr) 

^ H^"^ {ENgC X p*cEWgC -> p*cEWgC; Kr) . (3.17) 



25 



Since the projection EWqC x qc*EQc -^ EWqC is a VFcC-homotopy equiv- 
alence, the desired isomorphism now follows from (|3.17p . 

(ii) By Lemma IXTl 

Hf'^^{ENGC X pUEWgC X qd*Qc) ^ pUEWgC x qU*Qc);KR) 

- Hf''^ {ENgC X p*c (WgC X ^/c res^^c EWgC) 

= Hf «^ (ngCxa (res^^c ^iVcC x p^ res^^^^ EWgC)) 

^ iVcC x^ {p*c res^^fc £^W^gC) ; K^) . (3.18) 



The generator t of Qg — Z/2 acts on 

Hf^^{ENGC X p*g{EWgC x qd*Qc) ^ pUEWgC x qd*Qc); Kr) 
by the square of iVcC-maps 

ENgC X p*c{EWgC X qd*Qc) > p*c{EWgC x W*Qc) 

idxpc(idxrt) p^(id xrt) 

ENgC X p*g{EWgC x qd*Qc) > pUEWgC x qd*Qc) 

where rt : Qc — ^ Qc is right multiplication with t. 

To simplify notation, let X = res"^^^ E_NgC and Y = p^ leSy^^Q EWgC. 
Choose an element 7 S NgC that is mapped to i by gc : NgC -^ Qg, and equip 



H^'^^ {NgC XAiX xY)^ NgC xaY;K 



R 



with the Qc-action coming from square of A^cC-maps 

NgC xa {X X Y) > NgC xa Y 

VyXl -I r^xl - 

NgC Xa {X x Y) > NgC xa Y 

where l^-i is given by {x,y) n- (7^"'^ ■ a;,pc(7~^) • y) for x £ X and y <E Y 
and r~f is right multiplication by 7. It is straightforward to check that the 
isomorphism p.lSp is compatible with this Qc-action. 

Recall that Qg acts freely on A away from the identity. This implies that if 
t G Qc is the generator and a G A, then i • (a + to) = a + ta, and so a + ta = 0. 
Therefore the Qc-action on A is given by — id^i : A -i- A. Since the action of t 
on A is defined by conjugation by 7"^, the following diagram commutes. 

NgC — -4 NgC 



A ''"" > A 



26 



For any A-CW complex Z, this yields the commutative square 



H^''^{NgCxaZ;Kj,) 



f^oind , -I, 



ind 



"gC 



Hf{Z;KR) 



ind_ 



>i/f'=^(7VGCx^(Ax_id., ^);K, 



-^Hf{Ax^id^Z;KR), 



where ind^'^ denotes induction with respect to inclusion, and /* is induced by 
the map / : NqC ^^(^-i) {NqC x^ Z) — > NqC x^ {A x_id^ Z), which sends 
(fc, {g, z) to (fcc(7~^)((jf), (e, 0)). From the axioms of an induction structure, we 
also have that indc(7-i) is induced by the map /2 : NqC Xa Z -^ NqC Xc[-y~i) 
{NqC XaZ), which sends {k,z) to (e, (7/0, z)) (see [26l Section 1]). 

When Z = X X y or Z = F, we can define a map I : A x _ ^^^ Z —i' Z such 
that l{a,z) — aj^^z. Since ind^'^ is natural, this produces the commutative 
square 



H^^'^iNaC XA (A x_id^ Z);Kr) 



Hf{Ax_id^Z;Kf 



^Hf'''^{NGCxAZ--Kii) 



-^Hf{Z;KR), 



where I' : NqC Xa {A x_id^ Z) — ^ NqC xa Z maps (k{a,z)) to (fc, 07 ^z). 
Notice that the composition l'^ o f^ o indc(-y-i) — I'^o f*o (/2), is precisely the 
map r^ x l^-i used to define the Qc-action on H^°'~^ [NqC x^ (-'^ x F) -^ 
NqC xaY; K.r). Thus, the equivalence 



Ef'^^iNcC XAiX xY)^ NgC xaY; K^) ^ 



"gC 



Hf{XxY ^Y:Kr) 



is compatible with the Q^-action, where the action on HMX xY ^ Y^'Kr^ 
is induced by the composition ?* o ind_id^. 

Notice that X is a model for EA and F is a model for ED. Therefore, the 
map I defined above is unique up to A-homotopy. Since C C A is maximal 
cychc, there is a D C A such that C ® D = A and D = Z''"^ Thus, the 
identification 



Hf{X x y ^y;K/ 



= H'. 



CS)D 



{{EC X ED) X ED ^ ED; Kj 



Hf®°{EC xED^ ED; ¥.r) 



agrees with the Qc-action since the action on H^®^ {EC x ED — > ED; K^) is 

induced by {IcJd)* ° iiid(_idc e-idi,)! '^tiere Ic '■ ind_idc-EC — )► EC denotes 
the unique C-equivariant map (up to C-homotopy) and Id '■ ind_ id£, ED — > ED 



27 



denotes the unique D-equivariant map (up to D-homotopy). Note that this 
establishes the first desired isomorphism of Z[Qc']-inodules: 

Hfo^{ENGC X pUEWgC X qd*Qc) ^ pUEWgC x q^*Qc)\ K^) 



^ Hf{EA -^ p*cE{A/C); K 



Using the freeness of the action of D, 

jjCBD i^^fj X ED -^ ED; K^) = H^ {EC x BD ^ BD; Kr) . 
As in (13.31) . since BD is the {d — l)-dimensional torus, 

t^ (-^-^^ 

Hf{BDxEC^BD;KR) ^ iff_;(£;C ^ {.}; K^)^ ' ^ 

(=0 
1=0 

By Remark \TW[ the Qc-action on H^ {EC -^ {•};Ku), induced by (/c)* o 
ind_idcj corresponds to interchanging the two copies of NKf{R) in the decom- 
position 

H^{EC ^ {•}; Kr) ^ NK^R) © NK^R). 
Thus, we obtain an isomorphism of Z[(5c]-modules 

^{Z[Qc]®zNK,^i{R)y ' > ^ H^{BDxEC^BD;Kr), 

1=0 

which yields the desired result. D 

Proof of Theorem \1.12\ 

(i) Since i? is a regular ring, NKj{R) — {0} for every j G Z. Therefore, 

0Wh„(i^;i?)=Wh„(G;i?) 

-FGJ 

for every n G Z by Theorem 11.111 Furthermore, Kn{R) = for every n < —1. 
Thus, for any group F, the Atiyah-Hirzebruch spectral sequence implies that 

i7„(BF;K(i?)) =0 forn<-l, 

and the edge homomorphism induces an isomorphism 

Ho{Br;K{R)) ^ Ko{R). 

Hence, 

Wh„(F; R) ^ Kn{RT) for n < -1 

and 

Who(F;E) = coker(ii:o(i?) ^ Ko{RT)). 



(ii)]By Carter [TT], 

Kn{RF) ^ {0} for n < -2 



for every F ^ J. Now apply assertion (i) to complete the proof. D 



28 



3.3.2 i-theory in the case of a free conjugation action 

Proof of Theorem \1.13[ As in the iiT-theory proof, Theorem l2.1l and (|2.9p imply: 

Sr^^—'\G;R) := H^ {EG ^ {•};L^^°"^) 

= H^{EG -^ ^G;LJ^"°°^) ® H^{EG -^ EG; Lfl°°^)- 
From the G-pushout (|5^ . 

FeJ FeJ 

BydS^, 

0i7„^-^(^7VGG->p^^VFGG;4:°°>) - i/,?(^G^^G;4-°°>). 

Recall from the proof of Theorem 11.111 that I = Ii 11 /2 • 

If G e /i, then iVcG = A ^ G © Z''-! and W^gC = Z'^-i. Therefore, as 
in (1331), H^^^{ENgC -^ PcEWgC; L^~°°^) is isomorphic to: 

d-l 

F^®^'"(i?Gxra'^-i^ii;Z'^-i;4-°">) - 0iJ^_,(i?G^{.};4-°°>) - 

i=0 

by Remark 12.131 

Now assume G G /2- We obtain a M^gG- homology theory by assigning to a 
WcG-GM^-complex Z the Z-graded abelian group 

H^-^{ENgC X p*cZ ^ ptZ;L<,-°°>). 

Consider any free VFg G-G VF-complex Y . Then there is an equivariant Atiyah- 
Hirzebruch spectral sequence converging to 

Hlf{ENGC X p*cY ~> fcY; L^f °"^ 
whose i!^^-term is given by 

Elj = i?r^'' (r; Hf^''{ENGC x pJ^M^gG ^ p^W'gG; L<,-°°>)) . 
Using Lemma l377l 

Hf^^{ENGC X p*cWgC -^ p*cWgC- 4"°°^) 

^ Hf°^{NGC xcres^^cENGC -^ NgC xc {.};4"°"^) 
= iJf(res^^G^A^GG^{.};4-°°>) 
- ijf(i?G->{.};4:°°>). 

29 



But this is zero by Remark 12.131 Hence, £"? = for alH,j £ Z. This imphes 

that for every free WaC-CW-comple^ Y, H^'^'^ (ENgC xp*(jY -^ p*(jY] LJ^"°°>) 
vanishes. In particular, 

H^^'^iENaC X p*cEWgC ^ pJ,SW^gC;L<,-°°>) = 

and 

H^^^{ENgC X p*c{WgC xh EH) ^ p*c{WgC xh EH); L^7°°^) = 0. 

Therefore the Mayer- Vietoris sequence associated to the VFg C-pushout (|3.10p 
yields an isomorphism 

H^''''{ENgC X p*cWgC/H ^ p*cWgC/H; L^-""^) 
He.Jc 

^ i/f ^^ {ENgC X p*cEWgC ^ p*cEWgC; LJ,-°°>) . 

Since p'^ (H) ^ I?oo, Lemma 13.71 implies: 

HI^'^''{ENgC X p*cWgC/H ^ p*gWgC/H; LJ,-°°>) 

= H::-^{NgC x-..^. resgf ) ^TV^C ^ NgC X-^,^. {.}; L^°°>^ 



{- 
R 



- iJ^'(^) (resg<^^) ENgC ^ {.}; L<-°°> 



(- 

R 



= i?,f-(Si?oo^{.};L<-°°> 



- Vm\l-°°HD^;R). 

The projection E_NgC x PgEWgC —J- E_NgC is a -/VcC-homotopy equivalence, 
thus 

UNili— >(Z?^;i?) ^ H^^oC^^^^c^phEWGC^L^R"^^). 
HeJc 



Therefore, 






which implies that 

(0'5r'<"°°Hi^;i?))©(0 UNili— >(i?o,;i?)) ^5r'<~°°^(G;i?). 
\fgj ) \cei2 HeJc I 

Theorem 11.131 now follows from the fact that I2 is empty if Q has odd order. D 



30 



3.3.3 JC-theory in the case Q = Z/p for a prime p and regular R 

Lemma 3.19. Let f : G ^>- G' he a group homomorphism, T be a family of 
subgroups of G and R be a ring. Let X be a G' -CW -complex such that for every 
isotropy group H' C G' of X and every n g Z, 

Hfi"') {E:,nf^.^H'){r\H') ^ {.});Kn) - 0, 

where Tnf-^{H') := {K n f-^{H') \K eT}. Then, for every n G Z, 

H^{E^Gxf*X^f*X;KR)^0. 

Proof Sending X to H^{EjrG x f*X -> f*X\'Kn) defines a G'-liomology 
tlieory. Tliere is an equivariant version of tlie Atiyahi-Hirzebrucli spectral se- 
quence converging to Hf_^j {Ej^G x f*X -^ f*X; Kr) (see Davis-Liick [TH The- 
orem 4.7]). Its £^^-term is the Bredon homology of X 

El = Hf'^^^'\x,V,l 

where the coefficients are given by the covariant functor 

Vj : Or(G") ^ Z-MODULES 

defined by 

G'/H' ^ Hf{E^G X f*{G'/H') ^ /*(G7i/');Kfl). 

It suffices to show that the £'^-term is trivial for all i,j. We will do this by 
showing that for every subgroup H' C G" and every j £ Z, 

Hf{E^G X f*{G'/H') ^ f*{G'/H')-Ku) = 0. 

Using the induction structure and Lemma 13.71 

Hf{E^G X f*{G'/H') ^ f*{G'/H');KR) 

- Hf (g Xf-^^H') {res^:'^"'^ E^G ^ {•});K^) 

- Hf{GXf-^(H') {E^nf-HH'){r\H')) ^ {.})-\^r) 
= Hf'^"'^ {E:rnf-HH'){rHH') ^ {•});Kj,) , 

which is zero by assumption. D 

The following general lemma will also be needed. 

Lemma 3.20. Let Gi and G2 be two groups. Let X be a Gi-GW -complex, and 
let pr : Gi X G2 — ^ Gi be projection. Then, for every n G Z, there are natural 
isomorphisms 



( — 00' 



i/^^x'^^(pr*X;L^-°°>) - i/^,(^.L(— > 



31 



Proof. We use the notation from [211 Section 6]). Let pr: Or(G'i x G2) ^ 
Ot(G2) be the functor given by induction with pr. It sends (Gi x G2)/H to 
Gi/ pr(-ff ). Because of the adjunction between induction and restriction (see [TBI 
Lemma 1.9]) and }16[ Lemma 4.6]), it suffices to construct a weak equivalence 
of covariant Or(Gi)-spectra 



pr. 



(KflO^G.xG.^^K^^^^gG,^ 



By definition, pr,(Kflo5GixG2) gends d/H to Ki^(g<=i(Gi) x G^), where G^ 
is the groupoid with one object and G2 as its automorphism group. The desired 
weak equivalence of covariant Or(Gi)-spectra is now obtained by comparing the 
definition of the category of free R{Q'^^ (Gi) x G2)-modules with the definition 
of the category of free R[G2\Q'^^ (Gi)-modules. D 



Proof of Theorem \rU\ By Theorem [O and ^ 

Wh„ (G; R) ^ H^ {EG -^ EG; Kr) © H^ {EG -^ ^G; K^) . 

We begin by analyzing H^ {EG -^ EG; Kr) . 

Note that every non-trivial finite subgroup iJ of G is isomorphic to Z/p, 
and is maximal. Furthermore, the normalizer NqH is isomorphic to A^^p x H. 
This can be seen as follows. The homomorphism q: G ^ Q = Z/p induces an 
injection H — >■ Z/p, which must be an isomorphism for non-trivial H. Clearly 
j\z/p g^j^(j jj belong to NqH, and the subgroup generated by A^/^ and H is 
isomorphic to A^/t^ x H. Thus, it remains to show that NqH C A^/p x H. 
Let t € H he a, generator of H. Then every element in G is of the form at* for 
some i G {0, 1, 2, . . .p - 1} and some a e A. If at'' e NqH, then atH{at^)^'^ = 
ata~^ = P for some j e {0, 1, 2, . . .p-1}. Since q{t) = q{ata~^) = q{P) = q{ty , 
j — 1. Thus ata~^ = t and t^^at — a. This implies that a £ A^^p, and hence, 
at' e A^/P X H. 

From [SI, Corollary 2.10], there is a G-pushout 

Uhej G Xa^/pxh EA^/p X EH ^U EG 

Uh^jP 

UHejGxA-/p.HEA^/p >EG 

which induces an isomorphism 

H^{G y-A-f.>.H {EA^/P xEH)^G x^./,,^ EA^/P;Kr) 
He.i 

^H^{EG^EG;Kr), 
where J is a complete system of representatives of the conjugacy classes of 



32 



maximal finite subgroups of G. Since A '"p = Z^, the induction structure implies 
H^{G ^A-i.^H (EA^/P xEH)^G x^:,,,^^ EA^^P;Kn) 
= H^'' ''"{EA^/PxEH^EA^/P;Kr) 
^ H"{BA^/P X (EH ^ {.}); K^) 

= ^H^_,{EH^{.};Kn)^^ 

i=0 
e 

^ 0Wh„_,(i/;i?)O). 



Therefore, 



00Wh„_,(i/;i?)(') ^ H^{EG^EG;Kr). (3.21) 

He, J i=o 

Now we turn our attention to H^ {EG -^ ^G; K^). Let A := A/A^/p and 
G := G/A^^P. Then the exact sequence ()1.10p induces the exact sequence 

l^A^G^Q^l. (3.22) 

Let T be the family of subgroups X of G for which pr(i^) is finite, where 
pr : G ^> G is the quotient homomorphism. Let VCy be the family of virtually 
cyclic subgroups of G, VCyi be the family of virtually cyclic subgroups of type 
I, and Ti ^ T VCyi . The Farrell- Jones Conjecture in algebraic JiT-theory is 
true for any group appearing in J-, since every element in J- is virtually finitely 
generated abelian (see Theorem l2.ip . It is straightforward to check that pr* EG 
is a model for Ejr{G). By the Transitivity Principle (see |21[ Theorem A. 10] 
and [ini Theorem 65]) and Remark [HH 

H'r.iErAGY^B) ^ H^{w*EG;Kr). (3.23) 

Since every virtually cyclic subgroup of type I in G is infinite cyclic, every el- 
ement K £ VCyi belongs to Ti, or is infinite cyclic and {K' \ K' C K and K G 
Ti} consists of just the trivial group. Since R is assumed to be regular, the 
map Hf^{EZ;Kii) -^ iJ,f ({•}; Ki,;) is bijective for every n e Z. Therefore, the 
Transitivity Principle implies that 

H^{E^AG);Kr) - H^{EvcyAG);KR). (3.24) 

By Remark [m 

H^{EvcyAGy,KR) ^ H^{KG:.^r) (3.25) 

is also a bijection. Thus, p.23p . p.24p . p.25p and the Five-Lemma imply that, 
for every n G Z, 

H^{EG^EG;Kr) ^ H^{EG^pr*EG;KR). (3.26) 

33 



Next we show that the conjugation action of Z/p on A is free away from 
€ A. Let t be an element of G that is mapped hy q: G — > Z/p to a generator 
of Z/p. Let o e A be given such that pr(t) pr(a) pr(t)^^ = pr(a). Thus, there 
is a 6 e j4^/p such that tat^^ ^ ab. li p: A -^ A denotes the conjugation 
action with t, then p{a) — a = b in A, where the group operation in A is written 
additively. Since 

p— 1 p— 1 p—i 

p ■ b = Y,p\b) ^Y.p'i.pi''^ - «) - E^'^'(«) - '"'(«) = /'"w - « ^ 0' 

4=0 4=0 i=0 

it foUows that p(a) — a = 6 = 0. This imphes that a € A^/^, and thus a = in 
A. Therefore, the conjugation action of Z/p on A is free away from G A. 

Let J be a complete system of representatives of the conjugacy classes of 
maximal finite subgroups of G. From [30l Lemma 6.3] and [3H Corollary 2.11], 
there is a G-pushout 

U-Hel'^ ^H ^'^ —^ EG (3.27) 



U-HelG^Hi'} >EG 

Sending a G-GM^-complex X to the Z-graded abelian group Hf [EG x 
pr* X -^ pr*X;Ki^) defines a G-homology theory. Since pr^^({l}) = A^^^ = 
TJ^ and R is regular. Theorem 11.71 and Lemma 13.191 imply that if X is a free 
G-GM^-complex, then H^ (EG x pr*X ^ pr*X;Kfl,) = 0. Therefore, be- 
cause Ujj^-jG x-ffEH and EG are free G-GW-complexes, the Mayer- Vietoris 
sequence associated to the G-pushout p.27p yields an isomorphism 

H^ {EG X pr* G/H -^ pr* G/H; Kr) 
TleJ 

^ H^{EG X pr* EG -^ pr* EG; Kr) . (3.28) 

Notice that if pr^^(_ff) is torsion- free, then it is isomorphic to If . li H is 
trivial, then this follows from pr~^({l}) = A^^p. If _ff is a non-trivial finite 
subgroup of G, choose an element t e pT^^{H) such that pr(i) is a generator 
of H. Then every element in pr~^(7J) is of the form ai" for a G A^^p and 
u G {0, 1, 2, . . . ,p — 1}. For two such elements ai" and bt^ 

af^bt" = abf^t" = baff' = bfaf". 

Hence pr^^(iJ) is a torsion- free abelian group containing Z*^ as subgroup of finite 
index, and so pr~^{H) = Z'^. Thus, if pr^^(i/) is torsion-free, then Theorem 1 1.71 
and Lemma 13.191 imply that 

H^ {EG X pr* G/H -^ pr* G/H; Kr) = 0. 
34 



Therefore, if J is the subset of J consisting of those elements H £ J for which 
pr~^(7J) is not torsion-free, then (|3.28p becomes 

Hf^ {EG X pr* G/H -^ pr* G/H] K^) 



He.] 



Hn {EG X pr* EG -^ pr* EG; Kr) . (3.29) 



Let H e y. Then w~^{H) = A^/p x H. Since A^/p = Z^ the induction 
structure, Lemma 13.71 Lemma 13.201 and Theorem 11.71 imply 

Hn {EG X pr* G/H -^ pr* G/H; Kr) 

- H^(^Gx ^^1, ,Tj {vesf ^" EG ^ {•});Kr^ 

- Hf'xyesf'>^^EG^{.};KR) 

- Hf''x"{EA^/P^{.};Kn) 
^ Hr''{EA-/P^{.};K,,-^) 
^ WK{A^/P;R[I./p]) 

e-l 



0(iVK„-.(i?[Z/p]) © NKr,-.,{R[Z/p])) 

ceMxcy{A^/p) i=o 



C7^) 



Since the projection EG x pr* EG -^ EG is a G-homotopy equivalence, (|3.29p 
implies that 

e— 1 

^{NK,,^,{R[Z/p])(BNK,,^,iR[Z/p])f~^'^ 

TTej' ceMicyiA^/p) »=o 

^H^{EG-^pr*EG;K„). 

Together with p.26p . this produces the isomorphism 

[^{NK^^,{R[Z/p]) (B NK„^,{R[Z/p])fA 

Tiel' ceMicyiA^/p) \t=o / 

= H^{EG^EG;Kr). 

To complete the proof of the theorem, we must show that there is a bijection 
between the sets J and J . Send an element H £ J to the element in _ff ' G J that 
is uniquely determined by the property that pr(7f ) and H' are conjugated in G. 
This map is well-defined and injective because two non-trivial finite subgroups 
Hi and H2 of G are conjugate if and only if pr(iJi) and pr(iJ2) are conjugate 
in G. The map is surjective, since for any element H G J there exists a finite 
subgroup H C G with H — pr{H). This finishes the proof of Theorem 1 1.1 41 D 



35 



3.3.4 i-theory in the case Q = Z/p for an odd prime p 

Proof of Theorem 11.161 An argument completely analogous to the one that es- 
tablished p.21[) in the proof of Theorem 11.141 shows that 

^^S^-t^HH-Rp ^ H^{EG^EG;L^^-^). (3.30) 

HeJ i=0 

Since p is odd, every infinite virtually cyclic subgroup of G is of type I. Thus 
the result follows from Theorem 12. 1[ Remark 12.131 and (|3.30p . D 

Proof of Theorem \1.17\ There are natural maps L^ -^ h^ of covariant 
functors from GROUPOIDS to SPECTRA for j e {2, 1, 0, -1, . . .}. Let L^^^^'^^ 
be the covariant functor from GROUPOIDS to SPECTRA that assigns to a 
groupoid G the homotopy cofiber of the map of spectra obtained by evaluating 
h^' -^ Jj^ ' at G- We obtain equivariant homology theories -ffj(— ;Lg '), 

Hl{-;L^^^) and ijj(-; 4"''+^^^'^) (see Liick-Reich [29l Section 6]), such that for 
any group G and any G-CM^-complex X, there is a long exact sequence 

-^ H^_,{X;l.t'^) ^ H^_,{X-J.^^) ^ ... . (3.31) 



If X = {.}, then 



H^{{.}-J.t'^) - L</+i>(ZG); 

F^({.};4-'"+^^^">) = i/"(Z/2;X,(ZG)), 

and the sequence p.3ip can be identified with the Rothenberg sequence (|2.3p . 
Since L„ '(Z) ^ L„'(Z) is a bijection for every n e Z by (|2.3p . a spectral 
sequence argument shows that the map 

if„ [BG- L<^'+i> (Z)) ^ i7„ {BG- L<^> (Z)) 

is a bijection for every n e Z. Thus, H^ {EG;L^^^^'^'') ^ iJ„(BG; L<J+iJ>(Z)) 
is zero for all n G Z. Hence, for every n G Z, there is an isomorphism 

This implies that there is also a Rothenberg sequence for the structure sets 

> 5r'^<J'+i>(G;Z) ^ Sr'''^^'^(G;Z) ^ ff"(Z/2;^j(ZG)) 

^ 5P«.0-+i) (G; Z) ^ ^P"!^'^ (G; Z) ^ . . . . (3.32) 



36 



Since Mf„(Z[Z/p]) = for n < 1 (sec Bass-Murthy 0), Theorem Ol 
implies that, for every n < 1, 



00Wh„_,(iJ;Z)(°) ^ Wh„(G;Z). 



(3.33) 



HeJ i=0 



For every iJ e J and j < -2, Kj{ZH) = by Carter TT. Hence Kj{ZG) 
vanishes for j < —2 by (|3.33p . Theorem 11.161 and p. 321) imply that the map 

e 

He.] !=o 

is bijective for all n G Z and j G { — 1, —2, . . .} 11 {— cxa}. 

Next we use an inductive argument to show that this is also true for j = 0, 1. 
By taking the direct sum of the Rothenberg sequences (|3.32p for the various 
elements H E J and mapping it to p.32p for G, one obtains the following 
commutative diagram. 



eHeJm=oS^l^'\H;Z)i^ 



®HeJ ®Uo S^i^'^'^ (H; 7Lp , SrP'^ (G; Z) 



-^Sr'^'\G;Z) 



®HeJ ®to i?"-'(Z/2; K,{ZH)p ^ i7"(Z/2; K,{ZG)) 



®HeJ®USl'''tr\H;Zp 



®HeJ®USl-{^^\iH;I^P 



^Sl^'lp'^G-Z) 



^SI",^HG;Z) 



The middle horizontal arrow is an isomorphism since it is induced by the iso- 
morphism p.33p . which is compatible with the involutions. The Five-Lemma 



37 



implies that the maps 

HGJ i=0 

are bijective for ah n G Z if and only if the maps 

He.J i=0 

are bijective for all n £ Z. This takes care of the decorations j = 1 and 
j = 0. A similar argument, where Kj{ZH) and Kj{1jG) are replaced by Wh(iJ) 
and Wh(G) in the Rothenberg sequence, shows that this is also true for the 
decoration s, since it is true for the decoration h which is the the same as (1). 
This completes the proof of the theorem. D 

References 

[1] A. Alves and P. Ontaneda. A formula for the Whitehead group of a three- 
dimensional crystallographic group. Topology, 45(l):l-25, 2006. 

[2] M. Banagl and A. Ranicki. Generalized Arf invariants in algebraic L-theory. 
Adv. Math., 199(2):542-668, 2006. 

[3] A. Bartels. On the domain of the assembly map in algebraic iiT-theory. 
Algehr. Geom. TopoL, 3:1037-1050 (electronic), 2003. 

[4] A. Bartels and W. Liick. Isomorphism conjecture for homotopy /C-theory 
and groups acting on trees. J. Pure Appl. Algebra, 205(3):660-696, 2006. 

[5] A. Bartels and W. Liick. The Borel conjecture for hyperbolic and CAT(O)- 
groups. Ann. of Math. (2), 175:631-689,2012. 

[6] A. Bartels, W. Liick, and H. Reich. The if-theoretic Farrell- Jones conjec- 
ture for hyperbolic groups. Invent. Math., 172(l):29-70, 2008. 

[7] A. Bartels, W. Liick, and H. Reich. On the Farrell- Jones Conjecture and 
its applications. Journal of Topology, 1:57-86, 2008. 

[8] H. Bass and M. P. Murthy. Grothendieck groups and Picard groups of 
abclian group rings. Ann. of Math. (2), 86:16-73, 1967. 

[9] J. Bryant, S. Ferry, W. Mio, and S. Weinberger. Topology of homology 
manifolds. Ann. of Math. (2), 143(3):435-467, 1996. 

[10] S. E. Cappell. Unitary nilpotent groups and Hermitian if-thcory. L Bull. 
Amer. Math. Soc, 80:1117-1122, 1974. 



38 



[11] D. W. Carter. Lower ii'-theory of finite groups. Comm. Algebra^ 
8(20):1927-1937, 1980. 

[12] F. Connolly and A. Ranicki. On the calculation of UNil. Adv. Math., 
195(l):205-258, 2005. 

[13] F. X. Connolly and J. F. Davis. The surgery obstruction groups of the 
infinite dihedral group. Geom. TopoL, 8:1043-1078 (electronic), 2004. 

[14] J. F. Davis. Some remarks on Nil-groups in algebraic iiT-thcory. Preprint, 
arXiv:math.KT/0803.1641, 2008. 

[15] J. F. Davis, Q. Khan, and A. Ranicki. Algebraic iiT-thcory over the infinite 
dihedral group: an algebraic approach. Algebr. Geom. TopoL, 11(4):2391- 
2436, 2011. 

[16] J. F. Davis and W. Liick. Spaces over a category and assembly maps in 
isomorphism conjectures in K- and L-thcory. K -Theory, 15(3):201-252, 
1998. 

[17] J. F. Davis and W. Liick. The p-chain spectral sequence. K -Theory, 
30(1):71-104, 2003. Special issue in honor of Hyman Bass on his seven- 
tieth birthday. Part L 

[18] J. F. Davis and W. Liick. The topological K-iheory of certain crystal- 
lographic groups. Preprint, arXiv:1004.2660vl, to appear in Journal of 
Non-Commutative Geometry, 2010. 

[19] J. F. Davis, F. Quinn, and H. Reich. Algebraic X-theory over the infinite 
dihedral group: a controlled topology approach. J. TopoL, 4(3):505-528, 
2011. 

[20] W. Dwyer, M. Weiss, and B. Williams. A parametrized index theorem for 
the algebraic X-theory Euler class. Acta Math., 190(1):1-104, 2003. 

[21] F. T. Farrell and L. E. Jones. Isomorphism conjectures in algebraic K- 
theory. J. Amer. Math. Soc., 6(2):249-297, 1993. 

[22] F. T. Farrell and L. E. Jones. The lower algebraic X-theory of virtually 
infinite cychc groups. K-Theory, 9(l):13-30, 1995. 

[23] F. T. Farrell, L. E. Jones, and W. Liick. A caveat on the isomorphism 
conjecture in L-theory. Forum Math., 14(3):413-418, 2002. 

[24] M. Gromov. Hyperbolic groups. In Essays in group theory, pages 75-263. 
Springer- Verlag, New York, 1987. 

[25] J. Grunewald. The behavior of Nil-groups under localization and the rela- 
tive assembly map. Topology, 47(3):160-202, 2008. 



39 



[26] W. Liick. Chern characters for proper equivariant homology theories and 
apphcations to K- and L-theory. J. Reine Angew. Math., 543:193-234, 
2002. 

[27] W. Liick. K- and L-theory of the semi-direct product of the discrete 3- 
dimensional Heisenberg group by Z/4. Geom. TopoL, 9:1639-1676 (elec- 
tronic), 2005. 

[28] W. Liick. Survey on classifying spaces for families of subgroups. In Infinite 
groups: geometric, combinatorial and dynamical aspects, volume 248 of 
Progr. Math., pages 269-322. Birkhauser, Basel, 2005. 

[29] W. Liick and H. Reich. The Baum-Connes and the Farrell- Jones conjectures 
in K- and L-theory. In Handbook of K -theory. Vol. 1, 2, pages 703-842. 
Springer, Berlin, 2005. 

[30] W. Liick and R. Stamm. Computations of K- and L-thcory of cocompact 
planar groups. K-Theory, 21(3):249-292, 2000. 

[31] W. Liick and M. Weiermann. On the classifying space of the family of 
virtually cyclic subgroups. PAMQ, 8(2):497-555, 2012. 

[32] D. Mcintrup and T. Schick. A model for the universal space for proper 
actions of a hyperbolic group. New York J. Math., 8:1-7 (electronic), 2002. 

[33] K. Pearson. Algebraic iiT-theory of two-dimensional crystallographic 
groups. K-Theory, 14(3):265-280, 1998. 

[34] F. Quinn. Hyperelementary assembly for if-theory of virtually abelian 
groups. Preprint, arXiv:math.KT/0509294, 2005. 

[35] A. A. Ranicki. Exact sequences in the algebraic theory of surgery. Princeton 
University Press, Princeton, N.J., 1981. 

[36] A. A. Ranicki. Algebraic L-theory and topological manifolds. Cambridge 
University Press, Cambridge, 1992. 

[37] A. A. Ranicki. Lower K- and L-theory. Cambridge University Press, Cam- 
bridge, 1992. 

[38] J. Rosenberg. Algebraic K-theory and its applications. Springer- Verlag, 
New York, 1994. 

[39] T. torn Dieck. Transformation groups. Walter de Gruyter & Co., Berlin, 
1987. 



40